Enable coredns unit in workload cluster
What does this MR do and why?
As I described into issue #1548 (closed), into workload cluster fluentd is not able to send logs to loki in mgmt cluster because the endpoint ( loki.sylva) cannot be resolved. In order to solved this issue I've introduce coredns unit in workload cluster,in this way we are able to set the proper configmap for dns and pointing the right addres.
Today, the workload cluster is able to resolve endpoints by using the mgmt cluster DNS (using resolved service or resolv.conf configured via sylva-capi-cluster on each host). The aim of this MR is to configure coredns service on k8s level by collecting the external services from mgmt cluster and add them using hosts plugin in coredns on each workload cluster. In this way we can select exact those services which are needed to interact with the mgmt cluster and keep a proper config across the cluster.
Related reference(s)
Closes #1548 (closed)
Beside this MR another changes need to be completed:
-> in sylva-capi-cluster clean up old configuration related to dns settings ( sylva-projects/sylva-elements/helm-charts/sylva-capi-cluster!451 (merged))
-> in capi-rancher-import delete dnsPolicy and rely on default ( sylva-projects/sylva-elements/helm-charts/capi-rancher-import!46 (merged) )